Bad Apple or Rotten Tree? Institutional, Societal, and Military Determinants of Peacekeeping Abuses Marisella Rodriguez Brandon Kinne Department of Political Science University of California, Davis June 2, 2017 Under review at International Studies Quarterly Word count: 12,887 Abstract United Nations peacekeeping soldiers commit atrocities while deployed despite their purpose of protecting civilians from harm. Yet there is tremendous variance in the number of reported human rights abuses across peacekeeping missions. Why are some missions more susceptible to misconduct than others? To explain this variance, we identify three broad sources of influence on peacekeeping behavior, namely institutional, societal, and military determinants. We find that host and troop-contributing country institutions, particularly press freedoms, dramatically decrease violations. Contributing country compliance with international humanitarian law also decreases violations, though to a lesser degree than institutions, while societal influences have virtually no impact on abuses. We illustrate the utility of these findings by generating outof-sample predictions for hypothetical peacekeeping missions in countries with recent political turmoil. We conclude that the efficacy of peacekeeping missions depends in part on the strength of press freedoms in both host and contributing countries.
1
Introduction The United Nations (UN) establishes peacekeeping missions in order to quell conflict
environments, promote institutional capacity, and provide aid and protection to civilians. However, increased reports of soldier misconduct tarnish the reputation of these missions. The Associated Press released a damning report of criminal behavior, finding nearly 2,000 instances of abuse perpetrated by peacekeepers from 2004 to 2016 (Dodds 2017). Alarm over peacekeeping abuses dates to the early 1990s, where widespread sexual exploitation and abuse was discovered in Bosnia and Herzegovina, Kosovo, Cambodia, Timor-Leste, and, more recently, the Democratic Republic of the Congo and the Central African Republic (al Hussein 2005). Peacekeeping missions are also home to other behaviors, often criminal, that undermine faith in the mission, such as the transmission of harmful diseases, false detainment, and mass killing.1 However some missions are more successful in constraining soldier behavior than others. In 2006, the Office of Internal Oversight Services (OIOS) confirmed 176 allegations of abuse in the Democratic Republic of the Congo and no evidence of abuse in six other ongoing missions (United Nations General Assembly 2007).2 Why are some peacekeeping missions more susceptible to misconduct than others? In theory, we should expect limited variation in peacekeeping abuses across missions given the UN’s strict policy against the violation of civilian rights by personnel and soldiers. The UN developed a zero tolerance framework from the ambiguous regulation for staff to behave according to the “highest standards of...integrity” outlined in the 1948 Charter (United Nations 1945, Article 101). By 1999, the UN secretariat explicitly tied UN forces to 1
Reports suggest that peacekeeping soldiers are the cause of the cholera outbreak in Haiti (Bode 2016). The UN Mission in Kosovo issued compensation to two individuals who were imprisoned under false pretenses, a case we return to in further detail (Sweetser 2008). Finally, Human Rights Watch reports the murder of at least 18 individuals, including women and children, by peacekeeping soldiers in the Central African Republic as recently as 2016 (Human Rights Watch 2016). 2 Peacekeeping missions with no confirmed instances of abuse in 2006 include UNMOGIP, UNTSO, UNDOF, UNIFIL, UNAMA, and UNAMI (United Nations General Assembly 2007).
2
the principles of international humanitarian law and introduced new standards of conduct (Annan 1999).3 The increased clarity in identifying and responding to peacekeeping abuses should decrease a soldier’s motivation to abuse. Yet, reports continue to illuminate cases of gross misconduct. If the UN is an ineffective mechanism for soldier accountability, we must consider alternative constraints on soldier behavior. Against the backdrop of high-profile cases of peacekeeper abuse, many UN missions involve few reported abuses or no abuses at all. What explains this variation? Why are some missions seemingly afflicted with aberrant behavior while others manage to conduct their work with little negative impact on civilian populations? Drawing upon expansive literatures on human rights, inequality, military culture, humanitarian law, and UN peacekeeping practices, we identify three broad sources of influence on peacekeeper behavior. Institutional influences originate from the political, economic, legal, and juridical institutions of peacekeepers’ home countries, as well as from the institutions of the countries to which they are deployed. Countries with established rule of law are particularly well equipped to detain, prosecute, and punish human rights violators, for example, while countries with extensive press freedoms offer a more transparent informational environment where violations can be publicized and scrutinized. Societal influences originate from the larger societal characteristics of peacekeepers’ home countries. Countries with deep social inequalities, such as established gender hierarchies and patriarchal norms, socialize individuals into accepting the marginalization of large segments of society, which in turn predisposes those individuals toward dehumanization of vulnerable populations. Finally, military influences originate from the unique military cultures that constitute peacekeeping soldiers’ own professional back3
In response to mainstream media reporting on allegations of civilian abuse, the UN introduced the standards of conduct, “We are United Nations Peacekeeping Personnel,” and the personal code of conduct for soldiers, entitled “Ten Rules: Code of Personal Conduct for Blue Helmets.” The guidelines are unique in that they outline specific actions in violation with UN standards, such as “commit any act that could result in the physical, sexual or psychological harm or suffering to members of the local population,” “use unnecessary violence or threaten anyone in custody,” or “be abusive or uncivil to any member of the public” (United Nations General Assembly 2009, Annex H).
3
grounds. Militaries that actively promote compliance with international humanitarian law, for example, should produce soldiers that are especially sensitive to human rights abuses, while militaries that opt for volunteer recruitment over conscription should exhibit high levels of group cohesion and discipline. We do not ex ante favor one explanation over the others. Rather, we proceed inductively, extracting a series of testable hypotheses from the above trichotomous framework and conducting rigorous exploratory analysis. We test our hypotheses using newly collected data on peacekeeping abuses during the 2000–2010 period. The results show that societal influences—that is, the characteristics of the home countries from which peacekeepers originate—have virtually no impact on human rights abuses. In contrast, institutions like rule of law and, especially, press freedoms dramatically influence violations. Military culture, especially regarding compliance with international humanitarian law, also influences violations, though not quite to the same degree as institutions. We illustrate the utility of these findings by generating out-of-sample predictions for hypothetical peacekeeping missions. We select countries that have experienced recent political turmoil but no peacekeeping missions, such as Turkey and Colombia, and we show that as important institutional features like rule of law and press freedoms deteriorate over time, the risk of human rights abuses associated with a prospective peacekeeping mission spikes dramatically. We proceed in four sections. First, we discuss the literature on possible deterrents of human rights abuses specific to institutional, societal, and military constraints. We identify how rule of law, press freedoms, social norms, and military culture can impact soldier behavior. Second, we theorize how these existing mechanisms affect peacekeeping soldiers. We argue that troop-contributing and mission host countries play an institutional role in limiting abuses. However, there are mechanisms specific to contributors that constrain soldier behavior, such as societal norms promoting equality and the military’s adherence to international humanitarian law and recruitment practices. Third, we describe our data collection process 4
and research design, including three model specifications capturing institutional, societal, and military determinants of peacekeeping abuses. Fourth, we present our findings. We find that press freedoms for hosts and contributing countries have a strong and robust negative effect on peacekeeping abuses. To improve accountability, we recommend that host countries play a greater role in investigating and prosecuting perpetrators.
2
Existing Research on Peacekeeper Violations Figure 1 illustrates trends in human rights abuses for the period 2000–2010, disaggre-
gated to both the mission and country-host level. The data show substantial variation in outcomes. Some missions experience zero violations, while others experience hundreds. This variation is not well explained by traditional proxies like geography. For example, while the two most violation-prone missions—MONUC in the Democratic Republic of Congo and UNMIS in Sudan—are both in Africa, other missions hosted by African countries, such as in Chad or Angola, experienced relatively few violations. The puzzle, then, is clear. Why do some peacekeeping missions experience so many more human rights violations than others? Although the literature on peacekeeper abuses tends to focus narrowly on sexual exploitation and abuse (SEA), it nonetheless suggests helpful insights on the determinants of human rights abuses more broadly. We review the literature with the goal of emphasizing institutional, societal, and military influences on peacekeeper abuses.
Institutions. Institutionalized sources of accountability are essential to preventing peacekeeper abuses.
For example, Nord˚ as & Rustad (2013) find that SEA is less com-
mon in countries with high levels of economic development, consistent with the more general view that strong domestic institutions limit human rights violations (Beer & Mitchell 2004, Davenport 1999, Hamm 2001, Henderson 1991). In principle, the model UN
5
Figure 1: Human Rights Abuses by UN Peacekeepers, 2000–2010
200
Number of abuses
Number of abuses
300
200
100
100
Host country
U IN CA U M GU IN A U R IN CA U R S U O N U MIK N O U MI N G FI C Y U P N M I U N T U ME N M E IS ET M UN IN O U A R C U AT N M U IBH N TA U ET N D O U F N IF U IL N O U CI N T U SO N IK O M O U NU N A B M M IN SI U L M ST O N AH U SC U O N M U IL N A U M N AM I U ID N M M IS O N U C M
M
em
BO
N
.R
Ira q ep Su ub da .o n fC on go
0
D
ti
ia
ai H
be r Li
nd i Le on e
ra el
ru
Is
Bu
er ra
Si
W es
te
rn
Sa ha G ra ua te m al Yu go a C s en la vi tra a G lA eo fr i rg ca ia n R ep ub . C yp ru s Er itr ea An Bo go la sn ia −H C ha er d ze go vi na S Ea yria st Ti m Iv or or y C oa Le st ba no n
0
UN mission title
Note: Left panel shows abuses aggregated by country. Right panel shows abuses aggregated by peacekeeping mission. Vertical axis is number of unique abuse events. [NOTE: need to adjust figures to get x-axes aligned.]
Status of Forces Agreement (SOFA) for peacekeeping missions, which automatically takes effect until mission-specific agreements are drafted, establishes oversight of peacekeepers. However, the SOFA reserves criminal jurisdiction over peacekeeping soldiers for contributing states (Burke 2011). Further, as Notar (2006) notes, the SOFA does not contain strong assurances that contributing countries investigate and prosecute allegations of abuse. A functionally similar accountability mechanism can be activated via domestic rule of law. Defined most simply, rule of law is a system of transparent rights that allows individuals, including peacekeeping soldiers, to predict the legal consequences of their actions (Maravall & Przeworski 2003). Well-developed legal and juridical institutions ensure political rights, civil liberties, and mechanisms of accountability that are applicable to all citizens, regardless of political, social, economic, or military standing (O’Donnell 2004), all of which are crucial to preventing abuses. In addition to rule of law, press freedoms encourage oversight of public and private actors. A free media promotes accountability by publicizing flagrant violations of laws and
6
social norms. Livingston (1997) examines media presence according to intervention types, including peacekeeping missions, and argues that peacekeeping missions are susceptible to media attention in proportion to the host country’s degree of instability and fragility of peace. Because human rights abuses fundamentally conflict with the UN’s peacekeeping mandate and pose a vivid threat to peace, they are particularly high profile targets of media attention. Countries with established press freedoms allow the media to increase public awareness of peacekeeper abuses and facilitate prosecution. Overall, while the literature suggests that both rule of law and press freedoms help discourage abuses by peacekeepers, these possibilities have not, to our knowledge, been subjected to rigorous analysis.
Society. Societal norms may also play a role in peacekeeping missions. Peacekeeping soldiers are, first and foremost, citizens. They are thus susceptible to the influence of countryspecific social environments, which engender norms of behavior that, in turn, yield unique biases, preferences, and categorizations (Durkheim 1956, Campbell 1964, Neugarten, Moore & Lowe 1965, Carlson 2001, Horne 2003). McAdams (1997) describes norms as such non-legal obligations as beliefs, values, and social customs. Once norms become internalized, they are difficult to remove; Harro (2000) contends that socialization is a cyclical process, which begins before birth and is continuously reinforced over an individual’s lifetime. Peacekeepers act in accordance with prevailing societal norms by establishing internal sanctions for themselves, as well as external sanctions for others, when violating those norms (Olkinuora 1972). Societal norms represent a large, diffuse area of research, and the connection between a peacekeeper’s social environment and his subsequent behavior in a peacekeeping mission is likely complex. Nonetheless, the human rights literature has established a clear connection between inequality, broadly defined, and human rights violations (Landman & Larizza 2009, Melander 2005). With regard to peacekeeping, Gizelis (2009) finds that gender equality, in particular, increases the success of post-conflict peacekeeping operations and
7
reduces the probability of civil war recurrence. Karim & Beardsley (2016) similarly find that peacekeeping missions comprised of countries with strong records on gender equality are less likely to experience SEA allegations. However, the influence of gender equality on peacekeeper abuses more broadly—as well as the influence of other forms of inequality, such as economic or educational—remains unexplored.
Military. Institutions and social norms are general characteristics of peacekeepers’ respective home countries, but peacekeepers are also influenced by constraints and cultures unique to their militaries. An important influence on military culture is international humanitarian law (IHL). IHL binds all actors in a conflict to the laws of war, which include protections for civilians, noncombatants, and hors de combat (The Red Cross 2004, Askin 2003). Most mission mandates in fact reiterate IHL principles (Spencer 2005, 171). Yet, although peacekeeper abuses clearly fall under the protection of IHL, some scholars question the efficacy of international standards in limiting abuses (Notar 2006, Spencer 2005). Further, Morrow (2002) argues that governments employ laws of war strategically, in order to strengthen their own battlefield performance and prevent retaliatory violations by other states, which suggests that IHL does not exert a strong normative influence. Processes of military recruitment also affect military culture. States face a tradeoff between using tax revenues to pay for soldier wages in an all-volunteer force (AVF) or suffering the costs of potentially inadequate, unmotivated recruits in a conscription force (Ross 1994, 117). In terms of composition, conscription militaries are more likely to include wealthy or politically elite citizens uninvested in the cause, while AVFs more often consist of ideologically motivated, career-oriented soldiers (Vasquez 2005, Moskos 1970). Further, AVFs exhibit increased soldier morale and combat effectiveness, relative to conscript militaries, which suggests that group cohesion is higher in AVFs (Wong, Kolditz, Millen & Potter 2003, Gal & Manning 1987). The lower levels of group cohesion associated with conscript militaries
8
may have perverse consequences. For example, among rebel groups that employ involuntary recruitment practices, wartime rape has been adopted as a technique for group socialization; SEA enables rebel conscripts, who are often recruited by forced abduction and press-ganging, to create bonds of loyalty and trust (Cohen 2013, 461). Whether these motivations extend to peacekeeping soldiers is unclear. Although peacekeepers are, of course, distinct from rebels, many peacekeepers hail from conscript militaries, and peacekeeping missions themselves often suffer from poor group cohesion. And, as Wood (2009) notes, soldiers build ties not only through formal military training activities, but also through such informal activities as initiation rituals and hazing.
3
Institutional, Societal, and Military Influences on Peacekeeper Violations Drawing upon the above literature, we develop specific hypotheses regarding how in-
stitutional, societal, and military influences affect abuses by peacekeepers. For the sake of analytical clarity, we distinguish between attributes of mission host countries and troopcontributing countries. Importantly, some categories of influence apply to both host countries and contributors, while others apply only to contributors. Specifically, domestic institutions influence human rights abuses via both contributors and hosts, while societal and military influences are largely restricted to contributors. We discuss each category in turn.
3.1
Institutions: Rule of law and press freedoms
Effective rule of law ensures protection of political rights and civil liberties through clearly defined legislation and institutionalized accountability mechanisms (O’Donnell 2004). A domestic legal system specifies not only the rules of appropriate behavior but also the conse-
9
quences for violation of those rules. Transparent, publicly known accountability mechanisms allow civilians to predict the legal consequences of misconduct (Maravall & Przeworski 2003). For peacekeepers, rule of law matters at both the host country and contributing country levels. While the above-mentioned SOFA limits the jurisdiction of host countries over peacekeepers, a well-developed rule of law nonetheless allows hosts to promote accountability in at least three ways. First, host countries may play a role in victim redress. In principle, the UN investigates—and potentially provides compensation for—peacekeeper misconduct by way of a standing claims commission. Yet, although standing claims commissions are typically incorporated into SOFAs, they have never been invoked (United Nations General Assembly 1997, 4). Instead, third-party claims are commonly processed through informal “local claims review boards” or through negotiations with host countries (Dannenbaum 2010). Local claims review boards differ from standing claims commissions in that they maintain fewer “procedural safeguards”—for example, the UN maintains exclusive control over the reviewing panel and can refuse to disclose information to the public (Bode 2016, 772). Although their proceedings are private, local claims review boards do sometimes effectively impose accountability and provide compensation. Importantly, host-country legal systems often play a role in this process. For example, shortly after the Kosovo War in 1999, three brothers were falsely detained by personnel of the United Nations Mission in Kosovo (UNMIK) and were released only after a panel of international judges found no evidence for detainment (Sweetser 2008, Rawski 2002). The panel also ordered compensation for the Gashi brothers; importantly, the panel cited as a legal basis The Provisional Criminal Procedure Code of Kosovo, which states that victims of false detainment have the right to compensation (United Nations Mission in Kosovo 2003). Ultimately, the brothers received an undisclosed amount from a UNMIK claims commission—a direct result of the legal and administrative institutions already in place in Kosovo (Rawski 2002). Had it not been for Kosovo law identifying false detainment as a crime worthy of compensation, combined with 10
an established legal procedure for conducting investigations and issuing international rulings, the compensation may never have materialized. Second, host countries may file a complaint with the UN and negotiate for compensation on behalf of their citizens’ claims (Dannenbaum 2010). If a ruling finds in favor of the host country, the UN pays a lump sum, which the host country then distributes to victims (United Nations General Assembly 1996). For example, following misconduct against Belgian nationals during the UN Organization Mission in the Democratic Republic of the Congo (MONUC), the government of Belgium lobbied on behalf of its citizens’ claims of “damage to persons and property” (United Nations Security Council 1965, Annex I, 1). Nonetheless, successful instances of governments representing their citizens’ claims against the UN are rare, as mission host countries often lack the institutional capacity to bring collective claims against the UN. For example, Somalia neglected to lobby on behalf of their citizens’ claims, involving the UN Assistance Mission in Somalia (UNSOM), due to “the lack of an organized government in Mogadishu” at the time (Gibney, Tomaˇsevski & Vedsted-Hansen 1999, 280). Third, host countries promote accountability by arresting and detaining perpetrators. For example, the mandate for the UN Mission in Kosovo (UNMIK) established guidelines for an interim government (United Nations Security Council 1999), which in turn allowed the UNMIK civilian police force to train and work alongside the Kosovo Police Service (KPS). UNMIK and KPS collaborated to investigate a trafficking ring in 2006, ultimately arresting and detaining three UN police officers for human trafficking (United States Department of State 2007). In short, host countries retain a nontrivial ability to exercise “on the ground” restraint over peacekeepers. Nonetheless, host countries remain significantly limited in their capacity to hold soldiers legally accountable. Contributing countries exercise a more direct legal influence over peacekeepers than do hosts. As stipulated by SOFAs, contributing countries bear primary responsibility for applying appropriate legal standards to suspected human rights violators—as defined by their 11
domestic laws—and for subsequently bringing prosecution. Ceteris paribus, a country with a strong rule of law is better equipped to efficiently arrest, prosecute, and punish human rights violators. Clearly written statutes and transparent accountability mechanisms exemplify a government’s ability to effectively exercise its prosecutorial obligations. The Canadian government’s response to the “Somali Affair” offers insight into the potential investigative strength of a contributor country (Commission of Inquiry into the Deployment of Canadian Forces in Somalia 1997). Canada’s reputation as a peacekeeping contributor was tarnished in 1993 by the shooting of an unarmed Somali man and the torture and death of a Somali teenager (Trueheart 1993, Whitworth 2004). Canadian peacekeepers allegedly killed one victim execution style while the second was killed after “prolonged and severe pain and suffering” (Whitworth 2004, 91-92)(Commission of Inquiry into the Deployment of Canadian Forces in Somalia 1997). In response to criticism by a military surgeon, the Canadian government convened a commission to investigate the alleged misconduct and assess the overall efficacy of Canadian peacekeeping efforts (Commission of Inquiry into the Deployment of Canadian Forces in Somalia 1997, ES-2-4). The commission was expansive and transparent. An estimated “116 witnesses offered their evidence in open sessions broadcast on television across Canada” (Commission of Inquiry into the Deployment of Canadian Forces in Somalia 1997, ES-3). Both soldiers and officers testified on the events in Somalia and at the National Defence Headquarters (Commission of Inquiry into the Deployment of Canadian Forces in Somalia 1997, ES-3). The four-year inquiry led to significant findings. The commission uncovered allegations of excessive brutality and institutionalized racism, as well as a history of military leaders knowingly neglecting misconduct (Whitworth 2004, Commission of Inquiry into the Deployment of Canadian Forces in Somalia 1997). Though the inquiry was prematurely discontinued, the investigation led to the prosecution of nine Canadian peacekeepers (Commission of Inquiry into the Deployment of Canadian Forces in Somalia 1997, O’Brien 2012). 12
Both host and contributing countries offer mechanisms for accountability, to varying degrees. Host state institutions eliminate nontrivial barriers toward accountability on the ground by conducting thorough investigations and detaining perpetrators. Contributing state institutions, on the other hand, implement accountability by prosecuting soldiers at home. We hypothesize the following: Hypothesis 1 The stronger the rule of law in mission host countries, the lower the propensity for human rights abuses during peacekeeping missions. Hypothesis 2 The stronger the rule of law in troop-contributing countries, the lower the propensity for human rights abuses during peacekeeping missions. Media provides an additional source of accountability. As Puddephatt (2006) and Belknap (2002) argue, media plays a pivotal role in conflict by grabbing the attention of policymakers, the public, and international organizations alike. Indeed, by communicating the urgency of peacekeeper-induced human rights crises, the media can act as an important agenda setter. The extent to which the media can successfully set the agenda depends, in turn, on the extent to which a country guarantees basic press freedoms, such as protections on free expression, unimpeded access to media sources (digital, print, and broadcast), and transparent private ownership of media conglomerates. These freedoms allow the media to increase public awareness of peacekeeper abuses and, ceteris paribus, increase the likelihood that a given instance of abuse—and possibly the larger issue of peacekeeper abuses itself—receives a legal and/or criminal response. Media is a powerful tool for criminal prosecution by nurturing public participation. Media acts as a bridge between criminal investigators and the public by soliciting the public for information about perpetrators or violations witnessed (Ericson 1991). However, mass media also fosters passive public participation. Mass media is seen as the public’s primary source for understanding crime (Ericson 1991). As a result, news outlets shape viewers’ perception of 13
criminal and deviant behavior while simultaneously suppressing diverging opinions (Jewkes 2004). Regardless of a mission’s success, the media’s portrayal of abuses can negatively influence public opinion of peacekeepers and harm the efficacy of missions. The media also has an incentive to sensationalize events (Jewkes 2004).4 By distorting news content and crafting a “moral panic,” news outlets transform abuses into captivating narratives (Jewkes 2004).5 These narratives can induce public interest in the successful detainment and prosecution of peacekeepers. Given the public’s potential to help facilitate accountability, peacekeepers are deterred from committing abuses when press freedoms are strong. Importantly, press freedoms matter for both host and contributor countries. For contributors, a free press can monitor the activities of their country’s soldiers abroad, possibly through embedded journalists and other avenues of first-hand reporting. Further, by regularly reporting on high-profile cases, a contributor country’s media can publicize instances of abuse and inform the public of whether their government has adequately fulfilled its prosecutorial obligations. And because peacekeeping soldiers themselves remain under the legal jurisdiction of their home countries, they should be acutely aware that even journalistic accounts of human rights abuses can result in criminal prosecution. For hosts, a free press provides yet another neutral check on peacekeeper behavior. The presence of a free press should be readily apparent to peacekeepers themselves. When deployed to a country with high press freedoms, for example, peacekeepers are more likely to observe widely circulated forms of print and broadcast media, unimpeded access to digital media, a stronger presence for embedded journalists and war correspondents, and a government disinterested in media accounts of military misconduct. Further, even if peacekeeper violations fall outside the legal jurisdiction of the host country, reports of misconduct generated by the host’s media sources 4
Previous work shows that media over-represents physically violent criminal behavior and underrepresents petty crimes (Antunes & Hurley 1977, Sherizen 1978, Ditton & Duffy 1983). 5 Despite the potential for misinformation, research shows that the public largely accepts distorted news without question (Postman 1985, Edelman 1988).
14
can, in turn, percolate to international fora, including to the news outlets of contributor countries. The sexual assault of a Haitian teenager by four Uruguayan peacekeepers, which occurred during the United Nations Mission to Stabilize Haiti (MINUSTAH), provides an illuminating example. In 2011, a local journalist received cellphone video footage of peacekeeping soldiers collectively assaulting a Haitian teenager on the MINUSTAH base camp (Democracy Now 2011). The incident prompted widespread condemnation from Haitians and local media outlets, leading to a public demonstration by several hundred Haitians outside of the mission base (Ives 2013, Garces & Daniel 2011). Shortly after the news broke, Haitian President Michel Martelly released a statement “vigorously condemning” the assault, which was followed by a public apology from Uruguayan President Jose Muica and the repatriation of the perpetrators (Garces & Daniel 2011, Castaldi 2011). In 2012, the peacekeepers were charged and prosecuted in Uruguay, but with significantly lighter sentences than what they would have received under Haitian laws (Center for Economic and Policy Research 2012). After the Uruguayan press reported the outcome, protests renewed outside of the mission base in Haiti to demand the withdrawal of Uruguayan forces, which President Mujica agreed to the following year (Center for Economic and Policy Research 2012, Ives 2013). The media responses in both Haiti and Uruguay illustrate the ability of a free media to increase accountability for individual peacekeepers. News of the 2011 assault appeared in print, broadcast, and digital media, and it elicited critical responses from both the public and state leaders. Local and international condemnation heightened Uruguay’s sensitivity to the good standing of its peacekeepers, which in turn led to the prosecution, conviction, and punishment of individual violators, and the repatriation of all Uruguayan peacekeepers. This sensitivity to the power of the press should encourage soldiers to regulate their behavior in public and to carry that cautionary behavior to mission host countries. Hypothesis 3 The greater the freedom of press in troop-contributing countries, 15
the lower the propensity for human rights abuses during peacekeeping missions. Hypothesis 4 The greater the freedom of press in mission host countries, the lower the propensity for human rights abuses during peacekeeping missions.
3.2
Society: Gender and income inequality
The social environment in contributing countries further affects a peacekeeper’s motivation to abuse civilians. Countries with high levels of inequality socialize individuals to devalue certain groups or communities. Anecdotal evidence identifies oppressed communities as highly vulnerable to peacetime and wartime abuses (Martin 2005, Department of Economic and Social Affairs 2010, al Hussein 2005). Oppressed communities, such as women and children, lose their limited agency during conflict settings and are susceptible to the power of others—armed soldiers in particular. Further, evidence suggests that peacekeepers are most prone to abuse the same civilian groups abroad that they are socialized to devalue at home (Karim & Beardsley 2016). Combined, these various factors suggest not only that women are disproportionately targeted for human rights violations, but that the proclivity of peacekeepers toward targeting women depends, in part, on gender norms in contributing countries. Women’s participation in society and access to state institutions promotes egalitarian views (Hudson, Ballif-Spanvill, Caprioli & Emmett 2012, Karim & Beardsley 2016). These views translate into behavior as individuals are socialized to value women’s rights. Conceptually, a society’s adherence to principles of egalitarianism manifests in myriad ways, from women’s level of political participation to opportunities for gainful employment (Karim & Beardsley 2016, Melander 2005). Education plays an especially important role, given the close linkages between women’s education and such vital outcomes as economic development (Duflo 2012). In assessing the impact of patriarchal social influences on peacekeeping abuses, Karim & Beardsley (2016)
16
examine contributor countries’ ratio of girls to boys in primary schools, ultimately concluding that greater educational participation by women reduces incidents of SEA. We anticipate that the effect of gender equality in fact extends to all categories of peacekeeper abuses and is not limited to SEA. In contrast to Karim & Beardsley (2016), we emphasize gender parity in secondary school enrollment, as primary enrollment measures are sensitive to repetitive enrollments for over-age children (World Bank 2016a). As another important facet of gender equality, we also consider accessibility of healthcare. A woman’s access to healthcare is fundamental to her human rights and the social perception of her gender because it directly connects to her physical, emotional, and mental wellbeing (Otto 1995). Availability of healthcare during pregnancy and childbirth, in particular, helps identify the unique health standards adopted by a community and, further, indicates the extent to which a society values women. Ceteris paribus, social indicators like maternal mortality thus reveal crucial information about women’s overall social standing. In countries with high maternal mortality rates, patriarchal and inegalitarian values are more likely to be pervasive, which in turn translate into fewer normative constraints on the peacekeepers that hail from those countries and a greater risk of human rights abuses.6 Of course, inequality extends far beyond gender. Income inequality, for example, reflects an obdurate gap between the “haves” and the “have nots” (Landman & Larizza 2009). This gap creates incentives for the “haves” in society to “maintain control of their resources and to exclude access to those resources by the ‘have nots’ in ways that use coercive means that undermine the protection of personal integrity rights” (Landman 2013, 124). That is, the wealthy citizens of heavily unequal societies may willfully violate the personal integrity of poorer citizens in order to continue profiting from resources (Henderson 1991, Landman & 6
Maternal mortality rates are often used as a proxy for economic development (Hogan, Foreman, Naghavi, Ahn, Wang, Makela, Lopez, Lozano & Murray 2010). However, because our empirical models control separately for economic development, those aspects of maternal mortality that are uncorrelated with development should reflect social inequalities.
17
Larizza 2009, 719). Further, governments are often complicit in these violations. Ignoring abuse of the poor—or even promoting corruption, nepotism, and other policies that harm poor communities—allows governments and political leaders to actively restrict the distribution of resources (Landman & Larizza 2009). Peacekeepers exposed to social environments that promote inequality or reward elitism are more likely to hold negative views of those perceived as different, inferior, or undeserving. Hypothesis 5 The lower the gender parity of secondary enrollments in troopcontributing countries, the greater the propensity for human rights abuses during peacekeeping missions. Hypothesis 6 The greater the maternal mortality in troop-contributing countries, the greater the propensity for human rights abuses during peacekeeping missions. Hypothesis 7 The greater the income inequality in troop-contributing countries, the greater the propensity for human rights abuses during peacekeeping missions.
3.3
Military: Humanitarian law and conscription
Finally, we consider the constraining effects of military culture, focusing in particular on contributor countries’ adherence to IHL and their modes of military recruitment. UN Secretary-General Kofi Annan issued a bulletin on peacekeeping and IHL in 1999, informing member states that “the fundamental principles and rules of international humanitarian law” apply to UN forces when engaged as “combatants,” which Annan expanded to include “enforcement actions, or in peacekeeping operations when the use of force is permitted in self-defence” (Annan 1999). The consensus view of international legal scholars is that IHL generally applies to peacekeepers (Shraga 2000). Nonetheless, the influence of IHL on peacekeeper behavior does not necessarily hinge on the potentially controversial question of where 18
peacekeeping falls within IHL. Rather, we focus on the extent to which contributor countries have incorporated IHL into their domestic civilian and military legal structures. Soldiers who have been trained to observe IHL protections—and who have been informed of the consequences of IHL violations—are less likely to engage in abuses, irrespective of the abstract applicability of IHL to a particular mission. For example, the Deputy Force Commander of the peacekeeping mission to the former Yugoslavia (UNPROFO) professed that he witnessed multiple incidents “involving the abuse of basic human rights,” and that “a deplorable lack of knowledge of the basics of international humanitarian law [...] was preventing soldiers from taking positive action to stop or limit such abuses” (MacInnis 1996). While mission host countries can play an active role in holding peacekeepers accountable, contributing countries are ultimately responsible for instilling professional standards of conduct. Although IHL binds all parties to a conflict, states vary widely in their ratification and domestic incorporation of key treaties. We anticipate that greater efforts toward incorporation of IHL increase constraints on peacekeepers and, thus, reduce their propensity to commit abuses. As Morrow (2014) argues, states ratify binding humanitarian treaties in part to signal to their military personnel the need for restraint on the battlefield. Important steps that countries can take toward IHL compliance include ratification of the Additional Protocols to the Geneva Conventions (i.e., Protocol I and Protocol II), as well as the creation of national committees for the implementation of IHL. The purpose of a national committee is to “advise and assist governments in implementing and spreading knowledge of IHL,” which includes monitoring the application of IHL, proposing new legislation, and, most importantly for our purposes, assisting in the training of armed forces (International Committee of the Red Cross 2016b). National committees are not mandated by international law; rather, they are voluntary. As such, the creation of a committee reflects a particularly strong commitment to compliance. Finally, a country’s overall IHL-related legislative activity provides further evidence of compliance. Legislation involving fundamental protections—especially civilian 19
protections, such as rules on humane treatment, torture, and sexual violence—offers the greatest promise in stemming abuses. Hypothesis 8 The more strongly that troop-contributing countries adhere to international humanitarian law, the lower the propensity for human rights abuses during peacekeeping missions. Importantly, H8 poses a potential research design problem. Specifically, and paradoxically, IHL-related treaty making and legislative activity may be greatest among countries with established records of IHL violations. Frequent violators are the most likely to face international and/or domestic pressures to reform their IHL practices. At the same time, countries that perceive themselves as relatively compliant with IHL obligations often view treaties and implementing legislation as unnecessary—a position sometimes ascribed to the United States (Aldrich 1991). In short, IHL-related activities may proxy for noncompliant rather than compliant behavior, in which case efforts at compliance will correlate with more abuses rather than fewer. As described in the following section, we use a carefully specified research design to distinguish genuine efforts at IHL compliance from more superficial activities. A second influence on military professionalism is recruitment practices. Involuntary recruitment practices, such as state-endorsed conscription, negatively influence group cohesion and solidarity relative to AVF (Wong et al. 2003, Gal & Manning 1987). While compliance with directives is essential to achievement of military objectives, soldiers sometimes behave individualistically. Soldiers recruited through conscription, in particular, are more likely to lack a connection to the cause of conflict and a sense of loyalty to the mission (Osiel 1999, Gates 2002, Cohen 2013). In contrast, soldiers from AVF militaries are more likely to express solidarity with the group, to show professional respect for established rules of conduct, and to informally police the activities of their fellow soldiers. Ceteris paribus, 20
the greater the composition of AVF soldiers in a peacekeeping mission, the less likely human rights violations should be. A potential counterargument to this expectation—to which we return in the empirical analysis—is that because peacekeeping missions themselves are not “all volunteer” forces, the contributing countries’ method of recruitment is irrelevant or at least of lesser importance. If this counterargument holds, we would anticipate little or no effect for recruitment practices on abuses. Hypothesis 9 Involuntary recruitment in troop-contributing countries leads to a higher propensity for human rights abuses during peacekeeping missions.
4
Research Design To evaluate our hypotheses, we collected event data on peacekeeping abuses during the
period 2000–2010, inclusive. The online appendix thoroughly details our data collection procedures. Event data allow us to identify instances of abuse that might be unobservable from large-scale measures of mission efficiency or success (as provided by the UN or otherwise). The dependent variable is operationalized as the number of reported human rights abuses during a peacekeeping year. We identified 1,509 reported instances of peacekeeping soldiers abusing civilians during the period under consideration. When a report did not occur, we coded those mission years as no abuses. A potential danger of this approach is that abuses may be underreported, which may in turn bias the results of our analysis. This issue is especially relevant in estimating the effect of press freedoms on abuses, given that press freedoms may correlate with newspaper accounts of human rights violations. However, if such a bias exists, it in fact works against our hypotheses and raises the bar for our empirical tests. If news reports of abuses depend on press freedoms, then countries with strong press freedoms should see more reported abuses and countries with weak press freedoms should see fewer reported abuses. This expectation runs exactly counter to H3 and H4, which argue that 21
abuses are more likely, not less likely, in countries with poor press freedoms. We follow Karim & Beardsley (2016) in using a mission-year unit of analysis. For each observation, the dependent variable indicates the number of violations recorded in that peacekeeping mission in the current year. This unit of analysis allows us to incorporate attributes of both host countries and troop contributing countries. While data on host countries can be straightforwardly incorporated into the mission-year observations, data on contributors pose a level-of-analysis dilemma, as nearly all missions consist of multiple contributors. We again follow the approach of Karim & Beardsley (2016). For each contributor-country variable, we calculate a weighted average of that variable across all the countries in the mission, with the weighting for each respective contributor defined by that state’s annual average troop contribution to the mission. Because the number of observations is small, and because our operationalization of contributor-country variables requires careful attention to potential multicollinearity, we test our hypotheses using multiple specifications. We first specify an “institutional model,” which focuses on Hypotheses 1–4. For both hosts and contributors, we use the Freedom House data on press freedoms, which for each country and year assigns a score on a 101point scale, based on a combined analysis of each country’s legal, political, and economic environments (Freedom House 2016). We invert this scale so that larger values correspond to a freer press. To measure rule of law, we rely on the World Bank’s Worldwide Governance Indicators (WGI), which define rule of law as “perceptions of the extent to which agents have confidence in and abide by the rules of society, and in particular the quality of contract enforcement, property rights, the police, and the courts, as well as the likelihood of crime and violence” (World Bank 2016b). Second, we specify a “societal model,” which focuses on Hypotheses 5–7. To capture gender inequality in contributors, we include maternal mortality rates and secondary school enrollment rates distributed by gender (World Bank 2016a). Larger values of maternal 22
mortality should correspond to lower levels of development, while the converse is true of secondary enrollment. Since inequality takes several forms, we add an additional variable for income inequality, the Gini coefficient, which measures the extent to which a country’s distribution of income differs from a perfectly equal distribution (World Bank 2016a). We anticipate positive parameter estimates for the Gini coefficient and maternal mortality, and a negative estimate for secondary enrollment. Finally, we specify a “military culture” model, which tests H8 and H9. In order to assess the influence of IHL on peacekeeping abuses, we distinguish IHL legislation from efforts at IHL compliance. While legislation is of course a mechanism of compliance, legislation may also be a response to local and/or global pressures; that is, countries that pass large volumes of IHL-related legislation may simply be more prone to violations. For example, Sri Lanka has enacted voluminous legislation on fundamental guarantees, largely in response to international pressures, but has adopted none of the measures that we identify as necessary for enforcement. Legislation alone, without concrete implementation mechanisms, has little impact on compliance. We operationalize an IHL legislation index, focusing on six rules within customary IHL, as identified by International Committee of the Red Cross (2016a), that are particularly relevant to human rights abuses.7 To operationalize compliance in particular, we develop an index based on (1) whether governments have fully ratified Additional Protocols I and II of the Geneva Conventions, (2) whether governments have established national committees for the observance and enforcement of IHL, and (3) whether governments have implemented legislation on fundamental guarantees specifically at the military level, such as legislation involving military penal codes or military courts.8 By separating legislative 7
These are Rule 87, on humane treatment of civilians and hors de combat; Rule 139, on respect for humanitarian law; Rule 142, on providing instruction in IHL to armed forces; Rule 143, on teaching of IHL rights and responsibilities to civilians; Rule 149, on the state’s responsibility for IHL violations; and Rule 151, on criminal responsibility of individuals for war crimes. See International Committee of the Red Cross (2016a). 8 Data for both of these measures come from International Committee of the Red Cross (2016b), International Committee of the Red Cross (2016c), International Committee of the Red Cross (2016a).
23
activity from good-faith efforts at compliance, we hope to minimize the potential selection effects of IHL. To account for variation in military recruitment policies among contributors, we use conscription data from Asal, Conrad & Toronto (2015), updated through 2010. All three of the institutional, societal, and military models also include controls for logtransformed per-capita gross domestic product (GDP) (World Bank 2016a) and democracy (Boix, Miller & Rosato 2012), for both hosts and contributors. Because impoverished citizens are highly vulnerable to peacekeeping abuses (al Hussein 2005, Cs´aky 2008), controlling for economic development in host countries is crucial. Anecdotal evidence shows that citizens in need of food, water, and other basic human needs are more willing to exchange sexual favors for resources (Martin 2005, Cs´aky 2008). Citizens in extreme poverty are also vulnerable to prostitution and human trafficking (Kirby 2008). Peacekeeping abuse targeting impoverished citizens is further reinforced by a community’s dependence on the resources provided by peacekeeping operations, likely placing victims in closer contact with peacekeepers in general. For contributing countries, controlling for economic development allows us to distinguish between the effects of development and social inequalities. Finally, we control for the total number of peacekeeping troops within each mission-year, calculated as the sum of countries’ average annual contribution (Karim & Beardsley 2016, Nord˚ as & Rustad 2013) . Because these data are slightly skewed, we use a square-root transformation. Since a larger troop presence creates more opportunities for abuse, ceteris paribus, we expect the parameter estimate for this variable to be significantly positive. Empirically modeling peacekeeping abuses involves careful model specification. Because the count data are overdispersed, we employ a negative binomial model with standard errors clustered at the mission level. However, as we show later, our results are generally robust to alternative specifications, including population-averaged and random-effects models.9 9
We also considered zero-inflated negative binomial regression. A Vuong test indicated that the standard negative binomial functional form is preferable.
24
5
Empirical Results Table 1 presents the results of the negative binomial analysis, grouped into models that
separately assess institutional, societal, and military influences, respectively. The first column of Table 1 shows the estimates for the institutional model. We note first that, based on the model’s estimates, host-country rule of law is negatively and significantly related to violations. Countries with strong legal and juridical systems are, ceteris paribus, less likely to see human rights violations committed by UN peacekeepers within their borders. Interestingly, rule of law does not have the same effect on troop-contributing countries, as the estimate for contributors is indistinguishable from zero. This finding suggests that, although contributing countries have the legal capacity to prosecute peacekeepers, peacekeeping behavior is largely constrained by the legal standards in host countries. The UN Charter and Code of Conduct supports this finding by emphasizing the importance of respecting “all local laws and regulations” of host countries. However, the guidelines for peacekeeping behavior are unclear as to whether peacekeepers are held to their country of origin’s legal standards when abroad (United Nations General Assembly 1990, Article 6). At the same time, as noted earlier, contributor countries often face domestic resistance when prosecuting their own military personnel, which may further dampen the influence of contributor-country institutions. Overall, this set of results supports H1 but not H2. In contrast to rule of law, press freedoms matter at both the contributor and host levels. In short, as press freedoms increase in either hosts or contributors, counts of violations decrease. Although the magnitude of the effect is slightly larger for contributors than for hosts, the estimates are not significantly different from one another, which suggests that improvements in press freedoms at the host and contributor levels are equally likely to reduce violations. These findings strongly support both H3 and H4, and they reveal a potentially vital role for press freedoms.
25
Table 1: Negative binomial models, annual number of human rights abuses (1) Institutional Host country attributes Rule of law Press freedoms GDP/capita (logged) Democracy
-1.947∗∗∗ (0.434) -0.0458∗∗∗ (0.0155) 0.959∗∗∗ (0.220) 1.946∗∗∗ (0.565)
(2) Societal
(3) Military
0.105 (0.164) -0.460 (0.392)
0.243 (0.163) -0.622∗ (0.344)
Contributing country attributes (weighted) Rule of law Press freedoms
0.850 (1.007) -0.0743∗∗ (0.0296)
Maternal mortality
-0.00123 (0.00173) -0.0312 (0.0225) 0.0218 (0.0222)
Gini coefficient Secondary enrollment IHL legislation IHL enforcement Conscription -1.179∗ (0.697) GDP/capita (logged) 0.790 (0.575) 1 0.0308∗∗∗ Total troops 2 (0.00603) Constant -9.624∗∗ (4.268) ln(α) 0.674 (0.156) N 123 Pseudo R2 0.082 Log-likelihood -319.2 Standard errors in parentheses, clustered ∗ p < 0.1, ∗∗ p < 0.05, ∗∗∗ p < 0.01 Democracy
26
-2.380∗∗∗ (0.571) -0.552 (0.403) 0.0178∗∗∗ (0.00553) 5.960∗ (3.121) 0.846 (0.151) 123 0.053 -329.2 on mission
2.057∗∗∗ (0.466) -2.017∗∗∗ (0.461) 1.626 (1.025) -4.066∗∗∗ (0.774) 0.189 (0.254) 0.0139∗∗∗ (0.00483) 1.686 (1.750) 0.630 (0.141) 123 0.084 -318.3
Figure 2: Marginal Effects, Institutional Model
20 ●
10 ● ● ●
●
●
●
●
●
0
−10
30
20 ● ● ●
10
●
● ● ●
●
0 20th
30th
40th
50th
60th
70th
80th
90th
● ● ● ● ●
0
50th
60th
40th
50th
60th
70th
80th
70th
80th
●
90th
15
●
10
● ●
●
5
● ● ● ●
●
0
●
30th
40th
Host press freedom
●
20th
30th
Host rule of law
●
10th
20th
Percentile
●
10
10th
Percentile
30
20
Predicted annual number of abuses
30
10th
Predicted annual number of abuses
Contributor press freedom
Predicted annual number of abuses
Predicted annual number of abuses
Contributor rule of law
90th
10th
20th
Percentile
30th
40th
50th
60th
70th
80th
90th
Percentile
Note: Each figure shows marginal effect of increasing specified variable from the 10th to the 90th decile, holding all other variables at their respective means/medians. Blue lines are point estimates. Gray polygons are 95% confidence intervals.
The results for the control variables are somewhat surprising. For host countries, both per-capita GDP and democracy increase violations. However, these estimates are not robust across specifications. For contributors, per-capita GDP has no effect on abuses, while democracy, as expected, reduces the number of violations. Figure 2 plots the marginal effects for the key variables in Model 1. As suggested by the statistically insignificant estimate for contributors’ rule of law, the marginal effect of this variable is indistinguishable from zero. Press freedoms in contributor countries, on the other hand, have a nontrivial impact; as this variable increases from the 10th decile to the 90th, the predicted number of violations declines from about 15 per year to one per year. For hosts, rule of law reduces the predicted number of annual abuses from about 20 to virtually zero.
27
Figure 3: Marginal Effects, Societal Model Contributor maternal mortality
Contributor economic inequality
Contributor secondary enrollment 20
●
6
●
● ●
● ● ● ● ●
3
●
6
● ● ●
●
● ● ●
●
4
Predicted annual number of abuses
9
Predicted annual number of abuses
Predicted annual number of abuses
8 15
10
● ● ● ●
5
● ● ●
●
●
0 10th 20th 30th 40th 50th 60th 70th 80th 90th
10th 20th 30th 40th 50th 60th 70th 80th 90th
10th 20th 30th 40th 50th 60th 70th 80th 90th
Percentile
Percentile
Percentile
Note: Each figure shows marginal effect of increasing specified variable from its minimum to its maximum, holding all other variables at their respective means/medians. Blue lines are point estimates. Gray polygons are 95% confidence intervals.
The impact of press freedoms is also strong, reducing the predicted number of violations from about 12 to about one. As later analyses confirm, press freedoms are consistently one of the most robust predictors of violations. The second column of Table 1 shows the results of the societal model, which assesses the impact of peacekeepers’ domestic society on their proclivity to engage in human rights abuses. (As in all the models, we also control for both host’s and contributors’ per-capita GDP and regime type.) None of the estimates approaches statistical significance. These results suggest that the overall societal characteristics of contributor countries have little impact on the behavior of peacekeepers.10 Figure 3, which plots the marginal effects of the societal variables, further strengthens this conclusion. Contrary to 5–7, there appears to be no systematic relationship between societal factors and human rights abuses. The societal model addresses the broad context of peacekeepers’ home countries. As soldiers, however, peacekeepers are also influenced by military culture. The third column of 10 We also estimated separate models for each of Gini coefficient, secondary enrollment, and maternal mortality, and we explored alternative measures of development, including infant and child mortality, primary enrollment, and literacy. None of these measures yielded a consistently significant correlation with abuses.
28
Figure 4: Marginal Effects, Military Culture Model
40
30 ●
20
●
10 ●
Contributor conscription
Predicted annual number of abuses
Contributor IHL enforcement
Predicted annual number of abuses
Predicted annual number of abuses
Contributor IHL legislation
30
●
20
●
10 ● ● ● ●
●
●
●
●
●
●
10
●
●
5
●
●
● ●
●
●
● ●
● ●
0
0
15
10th 20th 30th 40th 50th 60th 70th 80th 90th
10th 20th 30th 40th 50th 60th 70th 80th 90th
10th 20th 30th 40th 50th 60th 70th 80th 90th
Percentile
Percentile
Percentile
Note: Each figure shows marginal effect of increasing specified variable from its minimum to its maximum, holding all other variables at their respective means/medians. Blue lines are point estimates. Gray polygons are 95% confidence intervals.
Table 1 shows the estimates for our proxies of military culture. IHL enforcement is strongly and negatively associated with human rights abuses, while IHL legislation is strongly and positively associated with abuses. This pair of results reinforces H8. When states make a concerted effort to enforce humanitarian law, not only by signing AP1 and AP2, but also by establishing national IHL committees and implementing legislation at the level of military penal codes, their peacekeepers are substantially less likely to commit violations. In contrast, when attempts at IHL compliance are merely legislative, peacekeepers are more rather than less likely to commit violations. This result requires a cautious interpretation. We do not believe that IHL legislation alone increases human rights abuses. Rather, this measure, in concert with the enforcement measure, proxies for countries that demand attention to IHL compliance issues—as reflected by large volumes of IHL-related legislation—but have not made significant progress toward enforcement. In addition to IHL, Model 3 reveals a positive correlation between conscription and human rights abuses, which supports 9. However, this estimate is not significant at conventional levels. Figure 4 plots the marginal effects from Model 3. The impact of IHL legislation in
29
contributor countries is strong and precise. Ceteris paribus, increasing this index from the 10th to the 90th decile correlates with an increase in predicted annual abuses from nearly zero to about 25. The impact of IHL enforcement is inverse to that of IHL legislation. As the value of the enforcement index increases, the predicted number of annual abuses drops precipitously from about 23 events to virtually zero. Conscription in contributor countries yields, at the 90th decile, a predicted count of nearly 10 abuses, though this estimate has wide confidence intervals. We next estimate a full model, combining all of the measures from the institutional, societal, and military models, as illustrated in Table 2. We urge caution in interpretation of these results, as the estimates may be sensitive to specification. Nonetheless, the results from the individual models are generally robust to the full specification. The signs and statistical significance of most estimates remain unchanged. The estimate for contributor democracy is an important exception. While this estimate was significantly negative in each of the three constituent models, it is now indistinguishable from zero. This result implies that influences commonly associated with democracy—such as press freedoms, rule of law, and IHL enforcement—bear the brunt of the work in reducing human rights violations; democracy alone is likely too broad and ambiguous of an influence. Table 2 also shows estimates from two additional model specifications. Model 5 is a population-averaged (PA) model with a negative binomial link function (Zorn 2001). The PA approach allows us to directly model within-group correlations, which are common in crosssectional time-series data. The mission-year counts of human rights violations are unlikely to be independent of one another. Instead, these counts vary systematically by mission, such that the number of violations in a given mission-year depends, in part, on already established patterns of violations within that mission. Specifically, we model the withingroup correlation structure as a first-order autoregressive process (AR1). The key results from the negative binomial model are robust to the PA specification. Rule of law and press 30
Table 2: Robustness checks, annual number of human rights abuses (4) Full model Host country attributes
(5) Full model
(6) Full model
-1.281∗∗ (0.574) -0.0506∗∗∗ (0.0152) 0.829∗∗∗ (0.265) 1.338∗∗ (0.587)
-1.128∗∗ (0.512) -0.0449∗∗∗ (0.0140) 0.226 (0.188) 0.531 (0.547)
-0.324 (0.452) -0.0334∗∗ (0.0132) 0.367∗ (0.192) -0.177 (0.479)
Rule of law Press freedoms GDP/capita (logged) Democracy
Contributing country attributes (weighted) Rule of law
-0.371 (1.062) Press freedoms -0.101∗∗ (0.0413) Maternal mortality 0.00331 (0.00254) Gini coefficient 0.0344 (0.0283) Secondary enrollment -0.00165 (0.0241) IHL legislation 1.269∗∗ (0.578) IHL enforcement -2.146∗∗∗ (0.507) Conscription 1.190 (1.095) Democracy -0.887 (1.250) GDP/capita (logged) 2.343∗∗∗ (0.847) 1 2 0.0222∗∗∗ Total troops (0.00584) Constant -17.87∗∗ (7.295) Method NegBin ln(α) 0.431∗∗ (0.168) N 123 Pseudo R2 0.111 Log-likelihood -308.9 Standard errors in parentheses; NegBin ∗ p < 0.1, ∗∗ p < 0.05, ∗∗∗ p < 0.01
31
-1.708∗ (0.940) -0.124∗∗∗ (0.0409) 0.00193 (0.00235) 0.0322 (0.0348) 0.0103 (0.0266) 0.584 (0.473) -1.975∗∗∗ (0.413) 1.880∗ (1.042) 0.178 (0.943) 2.459∗∗∗ (0.675) 0.0183∗∗∗ (0.00422) -14.23∗∗∗ (3.993) PA AR1
-1.312∗ (0.698) -0.103∗∗∗ (0.0375) 0.00539∗∗ (0.00248) -0.0130 (0.0255) 0.0311 (0.0218) 0.0590 (0.456) -1.515∗∗∗ (0.376) 0.314 (0.935) 0.581 (0.922) 2.110∗∗∗ (0.617) 0.00936∗∗ (0.00392) -14.66∗∗∗ (4.126) RE
111
123
-301.5 errors clustered on mission
freedoms are significantly negative for both hosts and contributors, and IHL enforcement is negative for contributors. IHL legislation loses its significance, but the parameter estimate for conscription is now positive and significant at the 10% level, supporting H9. Additionally, contributor rule of law is negative and significant at the 10% level. Finally, we estimated a random-effects (RE) model, which allows the dispersion in counts to vary randomly from one group to the next (while remaining the same for all observations in the same group). The RE model provides an alternative means of accounting for unobserved heterogeneity. Again, the key results from the negative binomial analysis generally hold up, with the most notable exception being host-country rule of law, the estimate for which is now statistically insignificant. However, press freedoms and IHL enforcement continue to be statistically significant, reinforcing the importance of these two influences on abuses. To illustrate the applicability of these results to real-world cases, we developed a novel out-of-sample predictive exercise. We identified crises that were considered by scholars and/or policymakers to be prime candidates for peacekeeping missions but, for various reasons, never received a mission. These “near missions” offer intriguing counterfactuals: if the missions had materialized, how many human rights abuses would we have likely seen? Combining the results of our model with different possible compositions of a hypothetical peacekeeping mission, we predicted the number of abuses for each mission. Our hypothetical missions include Spain as a host country in 2006, corresponding to ETA’s declaration of a permanent ceasefire; Ukraine in 2014, following the Minsk Protocol ceasefire; Syria in 2012, corresponding to the attempt at implementing Kofi Anna’s six-point peace plan in the Syrian civil war; Turkey in 2013, corresponding to the unilaterally declared ceasefire by the head of the Kurdistan Workers’ Party; and Colombia in 2015, following a ceasefire with FARC rebels, first proposed in 2012 and scheduled to take effect in 2016. These countryyears exhibit substantial variations in regime type, economic development, rule of law, press freedoms, and other key indicators. 32
Figure 5: Predicting Human Rights Abuses in Hypothetical Missions
Predicted number of abuses
150
Scenario
100
Best case Average case Worst case 50
0 Spain 2006
Ukraine 2014
Colombia 2015
Turkey 2013
Syria 2012
Hypothetical mission host country
Note: Bar heights indicate predicted number of abuses, based on Model 4, using “best case,” “worst case,” and “average case” values for contributing-country variables.
For each hypothetical mission, we set the host-country variables to their empirically observed values (using the same sources as referenced above), and we set the contributingcountry variables according to three distinct peacekeeping scenarios. In the “average case” scenario, we set the contributing-country variables to the observed means in the full sample of mission-years. In the “worst case” scenario, we set the contributing-country variables equal to the observed values from the 2006 Congo mission, which is the mission-year in our dataset with the highest recorded number of abuses. In the “best case” scenario, we use values from the peacekeeping mission in Georgia (specifically, the year 2005), which is notable for having no documented abuses despite a long tenure. Using the full model specified in the first column of Table 2, we generated out-of-sample predictions for each hypothetical mission, applying all three scenarios to each mission. Figure 5 illustrates the out-of-sample predicted abuse counts. The forecast for Spain in 2006 is the most optimistic, due to that country’s well-developed 33
Figure 6: Predicting Abuses in a Hypothetical Turkey Mission Observed attributes Percentage change since 2002
Predicted abuses (worst case)
Out−of−sample predictions 150
100
50
10
0
−10
−20
Press freedoms Rule of law
−30
Year
20 15
20 14
20 13
20 12
20 11
20 10
20 09
20 08
20 07
20 06
20 05
20 04
20 03
20 02
20 15
20 14
20 13
20 12
20 11
20 10
20 09
20 08
20 07
20 06
20 05
20 04
20 03
20 02
0
Year
Note: In left panel, bar heights indicate predicted number of abuses, based on Model 4, using “worst case” values for contributing-country variables. In right panel, lines show percentage deviations in press freedoms and rule of law from 2002 levels.
rule of law and press freedoms. Conversely, Syria in 2012 shows the worst projected outcomes, due to steady declines in both rule of law and press freedoms. The remaining three hypothetical missions fall between these two extremes. Overall, the results illustrate the joint importance of both host- and contributor-country attributes. For example, the worst case scenario for Spain, defined by a troop composition similar to that for Congo in 2006, yields a predicted number of abuses greater than the best case scenario for all other mission-years, including Syria in 2012. At the same time, even an unstable country like Syria yields relatively few predicted abuses if the mission is well composed; indeed, Syria’s best case scenario predicts fewer than ten abuses annually, only slightly more than Spain’s worst case scenario. Mission composition is clearly a central factor. At the same time, host-country attributes also dramatically influence outcomes. In Spain, a worst case mission would lead to fewer than ten abuses, while in Syria that same mission would lead to well over 150 abuses. Minimization of human rights violations requires careful attention to both host-country conditions and the attributes of contributors. To illustrate temporal dynamics of host-country attributes, Figures 6 and 7 focus on the
34
Figure 7: Predicting Abuses in a Hypothetical Colombia Mission Observed attributes Percentage change since 2002
100
50
Rule of law 30
20
10
Year
20 15
20 14
20 13
20 12
20 11
20 10
20 09
20 08
20 07
20 06
20 05
20 04
20 02
20 15
20 14
20 13
20 12
20 11
20 10
20 09
20 08
20 07
20 06
20 05
20 04
20 03
0 20 02
0
Press freedoms
40
20 03
Predicted abuses (worst case)
Out−of−sample predictions
Year
Note: In left panel, bar heights indicate predicted number of abuses, based on Model 4, using “worst case” values for contributing-country variables. In right panel, lines show percentage deviations in press freedoms and rule of law from 2002 levels.
cases of Turkey and Colombia, respectively. We used the same approach described above to predict the number of abuses for hypothetical peacekeeping missions in both countries, covering the period 2002–2015. We use the worst case scenario to predict the number of human rights violations annually. For reference, we also plot changes in host-country rule of law and press freedoms over this period. For Turkey, both rule of law and press freedoms improve slightly from 2002 through 2010, and, correspondingly, the predicted number of violations for a hypothetical peacekeeping mission generally remains below 50. From 2010 onward, however, the predicted number of abuses moves sharply upward, with a startling 163 abuses predicted for a hypothetical 2015 mission. The right-hand panel clearly shows that this spike is largely the result of a sharp downturn in press freedoms. This analysis reinforces the importance of timing. A peacekeeping mission in 2004 or 2005 would have, according to our model, resulted in no more than 30 annual human rights abuses, even in a worst case scenario. In Turkey’s 2015 political climate, however, that same mission would yield more than five times as many annual abuses. Figure 7 illustrates the more optimistic scenario where a country’s prospects improve
35
over time. From 2002 onward, Colombia’s rule of law improved by over 40% while its press freedoms improved by 20-30%. Accordingly, the projected number of abuses associated with a hypothetical peacekeeping mission declines by nearly half, from over 130 in 2002 to just over 70 in 2015.
6
Discussion & Conclusions We draw three main lessons from this analysis. First, there appears to be no “silver
bullet” for reducing human rights abuses by UN peacekeepers. Rather, abuses are a multidimensional phenomenon, driven by complex influences at the level of both hosts and contributors. This conclusion, though unsurprising, poses a challenge to peacekeeping missions, as it highlights the number of dimensions that the UN secretariat must consider when composing missions. Given the heterogeneity of states, and the fact that some contributors score well on some metrics and poorly on others, paying attention to these various dimensions is a nontrivial task. Given that missions struggle for contributors in the first place, we are not optimistic about the secretariat’s ability to assemble missions with this degree of attention to detail. Second, beyond this general lesson about complexity, a few influences stand out for their robust and nontrivial influence. Host-country rule of law, for example, consistently decreases human rights abuses. Institutionalized press freedoms exercise a particularly strong negative impact on abuses. Press freedoms are not only robust to model specification but in fact increase in both magnitude and precision as additional covariates are added to the model. We ran additional simulations of our full model in order to assess the impact of press freedoms on the missions illustrated in Figure 1. We found that increasing host and contributor press freedoms by just one standard deviation yielded a predicted ten abuses for the MONUC mission (compared to 277 observed abuses); slightly more than two abuses for UNMIS (com-
36
pared to 154 observed); and just 1.5 abuses for UNAMID (compared to 99 observed). Of course, press freedoms in host countries are rarely easily changed. Yet, contributor-country press freedoms matter just as much as host countries. Carefully assembling peacekeeping missions in order to emphasize press freedoms in contributors shows substantial promise of reducing abuses. Third, our findings suggest that host countries can help improve accountability of peacekeeping soldiers. The predicted effect of host country rule of law and press freedoms on abuses identifies an untapped potential for constraining misconduct. Peacekeepers are required to adhere to the local laws and regulations of host countries. However, host countries lack the legal capacity to prosecute violators. By failing to utilize host country institutions, the UN institutionalizes a relatively weak accountability structure. Of course, missions often occur when host country institutions are fragile. Yet anecdotal evidence suggests that host countries are capable of investigating and detaining peacekeepers. Giving host countries the freedom to enforce their respective legal standards against peacekeeping misconduct can help the UN realize their zero tolerance policy. Taken together, these lessons suggest that the efficacy of peacekeeping missions is improved by the institutional capacity of both host and contributing countries, particularly by strengthening press freedoms. Although missions depend on volunteer contingents, the UN’s reputation demands a more stringent selection process to prevent abuses. Additionally, the UN’s zero tolerance policy against misconduct is conditional upon expanding the legal capacity of host countries. In meeting these objectives, the UN can establish multiple modes of accountability to deter human rights abuses.
37
References al Hussein, Prince Zeid Ra’ad Zeid. 2005. A Comprehensive Strategy to Eliminate Future Sexual Exploitation and Abuse in United Nations Peacekeeping Operations, A/59/710. Technical report United Nations. Aldrich, George H. 1991. “Prospects for united states ratification of Additional Protocol I to the 1949 Geneva conventions.” American Journal of International Law 85(1):1–20. Annan, Kofi A. 1999. “Observance by United Nations forces of international humanitarian law.” International Review of the Red Cross 81(836):812–817. Antunes, George E. & Patricia A. Hurley. 1977. “The Representation of Criminal Events in Houston’s Two Daily Newspapers.” Journalism and Mass Communication Quarterly 55(4):756–760. Asal, Victor, Justin Conrad & Nathan Toronto. 2015. “I Want You! The Determinants of Military Conscription.” Journal of Conflict Resolution doi: 10.1177/0022002715606217. Askin, Kelly Dawn. 2003. “Prosecuting Wartime Rape and Other Gender- Related Crimes under International Law: Extraordinary Advances, Enduring Obstacles.” Berkeley Journal of International Law 21(2):288–349. Beer, Caroline & Neil J. Mitchell. 2004. “Democracy and Human Rights in the Mexican States: Elections or Social Capital?” International Studies Quarterly 48(2):293–312. Belknap, Margaret H. 2002. “The CNN Effect: Strategic Enabler or Operational Risk?” Parameters pp. 100–114. Bode, Thomas G. 2016. “Cholera in Haiti: United Nations Immunity and Accountability.” Georgetown Journal of International Law 47:759–791. Boix, Carles, Michael Miller & Sebastian Rosato. 2012. “A complete data set of political regimes, 1800–2007.” Comparative Political Studies 46(12):1523–1554. Burke, R´ois´ın. 2011. “Status of forces deployed on UN peacekeeping operations: Jurisdictional immunity.” Journal of Conflict and Security Law 16(1):63–104. Campbell, Ernest Q. 1964. “The Internalization of Moral Norms.” Sociometry 27(4):391–412. Carlson, Ann E. 2001. “Recycling Norms.” California Law Review 89(5):1231–1300. Castaldi, Malena. 2011. “Uruguay apologizes over alleged rape by UN peacekeepers.” Reuters . URL: http://www.reuters.com/article/us-haiti-uruguay-un-idUSTRE78603I20110907
38
Center for Economic and Policy Research. 2012. “Reduced charges against Uruguayan MINUSTAH troops latest example of lack of UN accountability.” Haiti: Relief and Reconstruction Watch . URL: http://cepr.net/blogs/haiti-relief-and-reconstruction-watch/reduced-chargesagainst-uruguayan-minustah-troops-latest-example-of-lack-of-un-accountability Cohen, Dara Kay. 2013. “Expaining rape during civil war: cross-national evidence (19802009).” American Political Science Review 107(3):461–477. Commission of Inquiry into the Deployment of Canadian Forces in Somalia. 1997. Dishonoured Legacy: The Lessons of the Somalia Affair. Technical report Canadian Government. Cs´aky, Corinna. 2008. No One To Turn To: The Under-reporting of Child Sexual Exploitation and Abuse by Aid Workers and Peacekeepers. Technical report Save the Children United Kingdom: . Dannenbaum, Tom. 2010. “Translating the Standard of Effective Control into a System of Effective Accountability: How Liability Should be Apportioned for Violations of Human Rights by Member State Troop Contingents Serving as United Nations Peacekeepers.” Harvard International Law Journal 51(1):113–192. Davenport, Christian. 1999. “Human Rights and the Democratic Proposition.” The Journal of Conflict Resolution 43(1):92–116. Democracy Now. 2011. “Video of UN peacekeepers’ sexual assault calls to focus on postquake rebuilding.”. Department of Economic and Social Affairs. 2010. The World’s Women 2010: Trends and Statistics. Technical report United Nations New York: . Ditton, James & James Duffy. 1983. “Bias in the Newspaper Reporting of Crime News.” British Journal of Criminology 23:159–165. Dodds, Paisley. 2017. “AP Exclusive: UN child sex ring left victims but no arrests.”. URL: https://apnews.com/e6ebc331460345c5abd4f57d77f535c1 Duflo, Esther. 2012. “Women empowerment and economic development.” Journal of Economic Literature 50(4):1051–1079. ´ Durkheim, Emile. 1956. Education and Sociology. Translated by Sherwood D. Fox Glencoe: The Free Press. Edelman, Murray. 1988. Constructing the Political Spoectacle. Chicago: University of Chicago Press.
39
Ericson, Richard V. 1991. “Mass Media, Crime, Law, and Justice: An Institutional Approach.” The British Journal of Criminology 31(3):219–249. Freedom House. 2016. “Freedom of the Press 2016.” https://freedomhouse.org/report/ freedom-press/freedom-press-2016. Gal, Reuven & Frederick J. Manning. 1987. “Morale and its Components: A Cross-National Comparison.” Journal of Applied Social Psychology 17(4):369–391. Garces, Raul O. & Trenton Daniel. 2011. “Haiti: Boy who claims sexual assault by Uruguay peacekeepers supported by demonstrators.”. URL: http://www.huffingtonpost.com/2011/09/06/haiti-sexual-assault-unpeacekeepers n 950159.html Gates, Scott. 2002. “Recruitment and Allegiance.” Journal of Conflict Resolution 46(1):111– 130. Gibney, Mark, Katarina Tomaˇsevski & Jens Vedsted-Hansen. 1999. “Transnational State Responsibility for Violations of Human Rights.” Harvard Human Rights Journal 12:267– 295. Gizelis, Theodora-Ismene. 2009. “Gender Empowerment and United Nations Peacebuilding.” Journal of Peace Research 46(4):505–523. Hamm, Brigitte I. 2001. “A Human Rights Approach to Development.” Human Rights Quarterly 23(4):1005–1031. Harro, Bobbie. 2000. The Cylce of Socialization. In Readings for Diversity and Social Justice, ed. Maurianne Adams, Warren J. Blumenfeld, Rosie Casta˜ neda, Heather W. Hackman, Madeline L. Peters & Ximena Z´ un ˜iga. First ed. New York: Routledge pp. 15–20. Henderson, Conway W. 1991. “Conditions Affecting the Use of Political Repression.” Journal of Conflict Resolution 35(1):120–142. Hogan, Margaret C, Kyle J Foreman, Mohsen Naghavi, Stephanie Y Ahn, Mengru Wang, Susanna M Makela, Alan D Lopez, Rafael Lozano & Christopher JL Murray. 2010. “Maternal mortality for 181 countries, 1980–2008: A systematic analysis of progress towards Millennium Development Goal 5.” The lancet 375(9726):1609–1623. Horne, Christine. 2003. “The Internal Enforcement of Norms.” European Sociological Review 19(4):335–343. Hudson, Valerie M., Bonnie Ballif-Spanvill, Mary Caprioli & Chad F. Emmett. 2012. Sex and World Peace. New York: Columbia University Press. Human Rights Watch. 2016. “Central African Republic: Murder by Peacekeepers.”.
40
International Committee of the Red Cross. 2016a. “Customary IHL Database.” https: //ihl-databases.icrc.org/customary-ihl/eng/docs/home. International Committee of the Red Cross. 2016b. “National Committees.” https://www.icrc.org/en/document/ table-national-committees-and-other-national-bodies-international-humanitarian-law. International Committee of the Red Cross. 2016c. “National Implementation Database.” https://ihl-databases.icrc.org/ihl-nat. Ives, Kim. 2013. “Uruguay will withdraw from MINUSTAH, President says.” Ha¨ıti Libert´e . URL: http://www.haiti-liberte.com/archives/volume7-16/Uruguay will withdraw.asp Jewkes, Yvonne. 2004. Media and Crime. California: SAGE Publications. Karim, Sabrina & Kyle Beardsley. 2016. “Explaining sexual exploitation and abuse in peacekeeping missions: The role of female peacekeepers and gender equality in contributing countries.” Journal of Peace Research 53(1):100–115. Kirby, Kristi M. 2008. “Peacekeepers, the Military and Human Traffiking: Protecting Whom?” University of St. Thomas Law Journal 6(1):221–246. Landman, Todd. 2013. Human Rights and Democracy: The Precarious Triumph of Ideals. New York: Bloomsbury. Landman, Todd & Marco Larizza. 2009. “Inequality and Human Rights: Who Controls What, When, and How.” International Studies Quarterly 53(3):715–736. Livingston, Steven. 1997. Clarifying the CNN effect: An Examination of Media Effects According to Type of Military Intervention. Technical Report June The Joan Shorenstein Center. MacInnis, John A. 1996. “Peacekeeping and international humanitarian law.” International Peacekeeping 3(3):92–97. Maravall, Jos´e Mar´ıa & Adam Przeworski, eds. 2003. Democracy and the Rule of Law. Cambridge: Cambridge University Press. Martin, Sarah. 2005. Must Boys Be Boys? Ending Sexual Exploitation and Abuse in UN Peacekeeping Missions. Technical report Refugees International. McAdams, Richard H. 1997. “The Origin, Development, and Regulation of Norms.” Michigan Law Review 96(2):338–433. Melander, Erik. 2005. “Political Gender Equality and State Human Rights Abuse.” Journal of Peace Research 42(2):149–166.
41
Morrow, James D. 2002. “The Laws of War, Common Conjectures, and Legal Systems in International Politics.” The Journal of Legal Studies 31(January 2002):S41–S60. Morrow, James D. 2014. Order within anarchy: The laws of war as an international institution. New York: Cambridge University Press. Moskos, Charles C. 1970. The American enlisted man: The rank and file in today’s military. New York: Russell Sage. Neugarten, Bernice L., Joan W. Moore & John C. Lowe. 1965. “Age Norms, Age Constraints, and Adult Socialization.” American Journal of Sociology 70(6):710–717. Nord˚ as, Ragnhild & Siri C. a. Rustad. 2013. “Sexual Exploitation and Abuse by Peacekeepers: Understanding Variation.” International Interactions 39(4):511–534. Notar, Susan A. 2006. “Peacekeepers as Perpetrators: Sexual Exploitation and Abuse of Women and Children in the Democratic Republic of the Congo.” American University Journal of Gender, Social Policy, and the Law 14(2):413–429. O’Brien, Melanie. 2012. “Protectors on Trial? Prosecuting peacekeepers for war crimes and crimes against humanity in the International Criminal Court.” International Journal of Law, Crime and Justice 40:223–241. O’Donnell, Guillermo A. 2004. “Why the Rule of Law Matters.” Journal of Democracy 15(4):32–46. Olkinuora, Erkki. 1972. “Socialization, Structure of Personal Norms, and Norm Alienation.” Acta Sociologica 15(3):228–253. Osiel, Mark J. 1999. Obeying Orders: Atrocity, Military Discipline, and the Law of War. New Jersey: Transaction Publishers. Otto, Dianne. 1995. “Linking Health and Human Rights: A Critical Legal Perspective.” Health and Human Rights 1(3):272–281. Postman, Neil. 1985. Amusing Ourselves to Death: Public Discourse in the Age of Show Business. New York: Viking. Puddephatt, Andrew. 2006. Voices of war: Conflict and the role of the media. Technical Report April International Media Support. Rawski, Frederick. 2002. “To Waive or Not To Waive: Immunity and Accountability in U.N. Peacekeeping Operations.” Connecticut Journal of International Law pp. 103–132. Ross, Thomas W. 1994. “Raising an Army: A Positive Theory of Military Recruitment.” The Journal of Law & Economics 37(1):109–131.
42
Schrodt, Philip A. 2012. “CAMEO: Conflict and Mediation Event Observations Event and Actor Codebook, Version 1.1b3.”. Sherizen, Sanford. 1978. Deviance and Mass Media. California: SAGE Publications chapter Social Creation of Crime News: All the News Fitted to Print, pp. 203–224. Shraga, Daphna. 2000. “UN peacekeeping operations: Applicability of international humanitarian law and responsibility for operations-related damage.” The American Journal of International Law 94(2):406–412. Spencer, Sarah W. 2005. “Making Peace: Preventing and Responding to Sexual Exploitation by United Nations Peacekeepers.” Journal of Public and International Affairs 16(Spring):167–181. Sweetser, Catherine E. 2008. “Providing Effective Remedies to Victims of Abuse by Peacekeeping Personnel.” New York University Law Review 83:1643–1678. The Red Cross. 2004. What is International Humanitarian Law? Advisory service International Committee of the Red Cross. Trueheart, Charles. 1993. “Canadian Peacekeepers Accused.” The Washington Post . United Nations. 1945. “Charter of the United Nations, 1 UNTS XVI.” United Nations. United Nations General Assembly. 1990. Comprehensive review of the whole question of peace-keeping operations in all their aspects. Technical Report A/45/594 United Nations. United Nations General Assembly. 1996. “Report of the Secretary-General, A/51/389.” United Nations. United Nations General Assembly. 1997. “Report of the Secretary-General, A/51/903.” United Nations. United Nations General Assembly. 2007. Special measures for the protection from sexual exploitation and sexual abuse. Technical Report A/61/957 United Nations. United Nations General Assembly. 2009. Manual on Policies and Procedures Concerning the Reimbursement and Control of Contingent-Owned Equipment of Troop/Police Contributors Participating in Peacekeeping Missions (COE Manual). A/c.5/63/18 ed. United Nations. United Nations Mission in Kosovo. 2003. “Provisional Criminal Procedure Code of Kosovo, UNMIK/REG/2003/26.” United Nations. United Nations Security Council. 1965. “Letter from the Secretary-General, S/6597.” United Nations. 43
United Nations Security Council. 1999. “Resolution 1244, S/RES/1244.” United Nations. United States Department of State. 2007. U.S. Department of State 2007 Trafficking in Persons Report - Serbia. Technical report United States Department of State. Vasquez, J. P. 2005. “Shouldering the Soldiering: Democracy, Conscription, and Military Casualties.” Journal of Conflict Resolution 49(6):849–873. Whitworth, Sandra. 2004. Men, Militarism & UN Peacekeeping: A Gendered Analysis. Bolder, Colorado: Lynne Reinner Publishers. Wong, Leonard, Thomas A. Kolditz, Raymond A. Millen & Terrence M. Potter. 2003. Why They Fight. Technical report US Army College. Wood, Elisabeth Jean. 2009. “Armed groups and sexual violence: when is wartime rape rare?” Politics and Society 37(1):131–162. World Bank. 2016a. “World Development Indicators, 1960–2015.” Accessed at http:// data.worldbank.org/data-catalog/world-development-indicators. World Bank. 2016b. “Worldwide Governance Indicators, 1996–2015.” Accessed at http: //info.worldbank.org/governance/wgi/index.aspx. Zorn, Christopher J. W. 2001. “Generalized estimating equation models for correlated data: A review with applications.” American Journal of Political Science pp. 470–490.
44
Appendix A: Data Collection on Peacekeeping Abuses We collected data on peacekeeping abuses using LexisNexis Academic and the Phoenix processing pipeline. Our objective was to collect precise information on peacekeeping soldiers violating the human rights of host country civilians during ongoing peacekeeping missions. This appendix details each step in the data collection process. We accessed LexisNexis Academic through the University of California, Davis library database to collect newspaper reports documenting instances of human rights abuses during peacekeeping missions. We searched LexisNexis content using a search string likely to capture instances of physical abuse by peacekeeping soldiers against host country civilians. For example, search string included terms such as “UN; united nations; peacekeep!; abuse; violence; assault; mission; troop*; soldier*”.11 We restricted the search command further to limit the timeline of events from January 2000 to May 2011. This search command allows up to six months for abuses in late 2010 to surface. The specific time period was chosen to comply with the Providing for Peacekeeping dataset. From there, we restricted search findings to newspaper articles. Newspaper articles are an advantageous resource for events data because they provide large amounts of information in a highly condensed delivery. Although official records, perhaps from the UN or participating countries, would likely provide more information, the reporting of peacekeeping abuses is highly sensitive information that, until recently, was not widely acknowledged at the institutional or state level.12 However, multiple newspaper outlets often report on a single event. To avoid duplication, we restricted the results further to exclude articles with high similarity.13 We then downloaded the newspaper 11
The search truncation (!) and wildcard (*) terms finds the root word along with a combination of additional letters. For example, “peacekeep!” searches for “peacekeeper” and “peacekeeping”. Two other search strings used are (1) “UN; united nations; peacekeep!; abuse; violence; assault; mission and (2) “UN; united nations; peacekeep!; abuse; violence; assault; mission; troops; soldiers”. 12 The UN’s Office of Internal Oversight Services has recently started to release reports evaluating mission effectiveness in deterring abuses, largely specific to the issue of sexual exploitation and abuse 13 We used the “Duplicates Option On – High similarity” command.
45
articles and uploaded them to the Phoenix processing pipeline.14 We used the Phoenix event data parsing platform provided by the Open Event Data Alliance (OEDA) to parse out the necessary information into data points. The raw text is cleaned and parsed using Stanford National Language Processing (NLP) text tree protocols. The Phoenix platform uses sentence structure and dictionary entries to assess initiator/target/action sequences out of sentence structures. Records are then de-duplicated by actor dyad, date, and event code, resulting in only one event code per actor dyad, per day. The data is then compiled according to the 20 broad event categories defined by the Conflict and Mediation Event Observations (CAMEO) based on the topic and structure.15 We focused on four subcategories under the general category, “Assault”. The four subcategories identify distinct forms of physical abuse, including acts of torture, the use of unconventional violence, sexual violence, and death by physical assault (Schrodt 2012). The subcategories are defined by Schrodt (2012, 81) as follows: • Torture: inflicting extreme pain on individuals. • Unconventional violence: physical violence which does not require high levels of organization or conventional weaponry. • Sexual violence: sexually abusing or assaulting the sexual integrity of individuals. • Kill by physical assault: killing individuals by physically assaulting them without the use of weaponry. Once we compiled the data into subcategories, we restricted the identifiable actors to peacekeeping soldiers (“IGOPKO”) and civilian victims (“CIV”). This condition ensures 14 LexisNexis allows for a maximum of 500 articles downloaded in a single batch. We then converted the articles into .txt files for ease of use when uploading the documents to the Phoenix parsing platform. 15 There are a total of 20 broad CAMEO event code categories and 283 subcategories. For a complete list of event code categories, see the CAMEO codebook.
46
that the data is representative of events where peacekeeping soldiers (Actor A) commits transgressions against host country civilians (Actor B). We then merged the event dataset with the Providing for Peacekeeping dataset, which includes information on mission-level characteristics from 2000 to 2010.16 The Providing for Peacekeeping dataset includes the location, time period, contributing countries, and number of peacekeepers for ongoing missions. We used this information to drop events that did not occur in host countries or during active mission years. Limiting the location and time period of possible abuses to match that of the Providing for Peacekeeping dataset ensures the plausibility of reported abuses during active deployment. From here, we proceeded to compile open sourced data on host and contributing country attributes.
16
We merged the two datasets using country year.
47
